PDA

View Full Version : Panoptes: A Binary Translation Framework for CUDA


News
05-22-12, 04:50 AM
Traditional CPU-based computing environments offer a variety of binary instrumentation frameworks. Instrumentation and analysis tools for GPU environments to date have been more limited. Panoptes (http://github.com/ckennelly/panoptes) is a binary instrumentation framework for CUDA that targets the GPU. By exploiting the GPU to run modified kernels, computationally-intensive programs can be run at the native parallelism of the device during analysis. To demonstrate its instrumentation capabilities, we currently implement a memory addressability and validity checker that targets CUDA programs.

Panoptes traces targeted programs by library interposition at runtime. Interactions with the GPU are intercepted, annotated as necessary, and are then sent to the actual CUDA library for execution on the device. This approach gives an analysis tool built on Panoptes a complete view of the state of the GPU without additional developer effort. In contrast, developer-added instrumentation may be incomplete due to errors of omission or cause maintenance difficulties, particularly for large code bases.

By directing annotated instructions to the GPU for execution rather than relying on the host for emulation, Panoptes is able to analyze programs at scale. The rift in parallel execution capabilities between modern GPUs and CPUs carries into testing and debugging as well. For computationally intensive tasks brought to the GPU explicitly for its parallelism, resorting to host-based emulation may necessitate reduced or simplified inputs for analysis. More details: http://github.com/ckennelly/panoptes



More... (http://gpgpu.org/2012/05/22/panoptes-a-binary-translation-framework-for-cuda)